Maximum network lifetime with optimal power/rate and routing trade-off for Wireless Multimedia Sensor Networks

Wireless Multimedia Sensor Networks (WMSN) are today considered as a promising technology, notably because of the availability of miniaturized multimedia hardware (e.g., CMOS cameras). Nevertheless, they do raise new research challenges; i.e., multimedia content is much more voluminous and rich in comparison to scalar one. Hence, multimedia data need pre-processing, namely coding, before they are delivered to the sink. In WMSN, meeting the required visual quality at the sink whilst taking into account the intrinsic network limitations, energy in particular, remains very challenging. In fact, higher visual quality of the data increases its volume, which in turn consumes more network resources and vice versa. The problem is more complex when including the data routing, especially when the routing is initially unknown. In this paper, we focus on the problem of simultaneously optimizing the video encoding at the source nodes and the routing of the generated data (i.e., the routing matrix is initially unknown) to the sink in order to maximize the network lifetime. This issue is tackled through the proposition of an analytic model. Based on the latter, two solutions have been studied. In the first one, of a static nature (i.e., network topology is static), the routes are calculated using the shortest path routing protocol toward the sink. The evaluation we made has shown that our proposed solution consumes less than 0.030% of the total battery while it increases the whole network lifetime nearly between 7 and 12 times, depending on the considered topologies, in comparison to the baseline approach (i.e., without optimization). Afterwards, in order to handle dynamic topology changes and to consider the link reliability, the work have been extended to the selection of routing paths, in a fully distributed fashion (i.e., using a local decision hop-by-hop routing protocol) with respect to both the shortest ones and their reliability. The second solution was evaluated through two different dynamic topologies. The simulation results show an energy consumption inferior to 0.025% and 0.080% of the total battery in the first and the second topology, respectively, with an improvement of the network lifetime that is multiplied by around 7 in comparison to the baseline approach.